Developing imidazoles as CEST MRI pH sensors

Abstract

A series of intra-molecular hydrogen bonded imidazoles and related heterocyclic compounds were screened for their N–H chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) contrast properties. Of the compounds, imidazole-4,5-dicarboxamides (I45DCs) were found to provide the strongest contrast, with the contrast produced at a large chemical shift from water (7.8 ppm) and strongly dependent on pH. We have tested several probes based on this scaffold, and demonstrated that these probes could be applied for in vivo detection of kidney pH after intravenous administration.

1. INTRODUCTION

Ischemia is a common cause of acute kidney injury, with biomarkers to detect this condition an active area of research (1). Acidosis is a factor in the extension of this injury (2), and as a result pH measurements represent one promising biomarker for detection (3). Compared with microelectrode measurements, less invasive magnetic resonance (MR) methods are attractive, taking advantage of the exquisite spatial and temporal resolution that these methods provide (4). Significant progress has been made in both magnetic resonance spectroscopy (MRS) (5–11) and relaxation based MR probes for in vivo pH detection (12–16). Recently, hyperpolarized ¹³C-labelled bicarbonate has also been applied to MR pH imaging (17). Chemical exchange saturation transfer (CEST) has been developed as a new method to improve the inherent sensitivity of MR (18–22). The CEST contrast mechanism involves selective saturation of labile protons on the CEST agent, followed by rapid proton exchange between these labile protons and bulk water. This process could offer several-thousand-fold signal amplification and several advantages over relaxation-based contrast agents, such as the capability of switching contrast between on and off (18–22), and simultaneous, multi-frequency detection (23–25). Several paramagnetic CEST (paraCEST) agents (26–32) and diamagnetic CEST (diaCEST) (33–41) agents have been developed as pH imaging probes. Some examples are shown in Fig. 1.

Figure 1.

Examples of paraCEST and diaCEST pH sensors.

A variety of compounds are attractive as diaCEST MRI contrast agents because of their bio-compatibility, including glucose (42,43), glycogen (44), glutamate (45), creatine (46), L-arginine (25), glycosaminoglycans (47), glycoproteins (48), nucleic acids (49–51) and peptides (24,52). However, a limitation in the sensitivity of most diaCEST agents is their small chemical shift from bulk water. Common metabolites show CEST contrast at around 1–3 ppm, which is improved to 5.5 ppm by barbituric acid (53) and thymidine analogues (50). Iodinated diaCEST/CT agents have been shown to provide exchangeable protons at about 6 ppm with a dependence on pH suitable for sensing (35,40,54), including for imaging acute kidney injury (3). As we have shown recently, it is possible to increase the shift above 6 ppm by employing the intra-molecular bond shifted hydrogens (IM-SHY principle). Previously we developed a library of diaCEST agents based on OH and NH protons in phenol and aniline derivatives (55–57). Here we were interested in designing pH sensors with large chemical shifts using IM-SHY. In particular, we report imidazole-4,5-dicarboxamides (I45DCs) as new pH sensitive diaCEST agents, which can provide CEST signal at 7.8 ppm from water with improved sensitivity and specificity compared with existing agents. In addition, we have evaluated one I45DC, Compound 5, to measure the pH of kidney in mice after intravenous (IV) administration.

2. RESULTS AND DISCUSSION

Heterocycles, including pyrrole, imidazole, pyrazole, imidazole, triazole, tetrazole and their analogues, represent another class of compounds where exchangeable protons could be provided at downfield chemical shifts due to the strong de-shielding effect from the aromatic ring and capacity for concurrent intramolecular hydrogen bonding (see below). Importantly, they are also capable of registering a response to changes in pH. For example, the chemical shift of the C-2 proton on certain N-alkylated imidazoles suggests 2-imidazole-1-yl-3-ethoxy-carbonyl propionic acid (IEPA) as an important MRS pH sensing agent (11). Inspired by this, we investigated the capacity of imidazole N–H exchangeable protons for pH sensing via CEST. However, imidazole (1, Fig. 2) gave poor saturation transfer efficiency and failed to give significant CEST contrast at neutral pH due to its fast exchange rate with water (k_aw).

Figure 2.

Azole derivatives analysed for their CEST properties.

Further inspiration for the compounds shown in Fig. 2 and studied here was derived from the catalytic site of serine proteases. Such proteases have a well-defined hydrogen bonding network, which slows the exchange of the imidazole N–H groups in histidine sufficiently to allow detection through CEST contrast, with the exchangeable proton demonstrating chemical shifts as far as 13 ppm downfield from water (58,59). Aiming to decrease the k_aw to the slow–intermediate exchange regime for our scanners to allow detection through CEST MRI, imidazoles were synthesized that might contain heterocyclic N–H functions as partners in an intra-molecular hydrogen bond (Fig. 2). After a series of negative results (2–4, and several other compounds listed in the Supporting Information), we found 4,5-bis[(Glu)carbonyl]-1H-imidazole (5), which provides significant contrast at 7.8 ppm downfield from water at neutral pH. Glutamate was chosen as the side chain to increase water solubility and reduce aggregation. The 1,2,3-1H-triazole analogue (6) failed to provide any CEST signal, even the expected amide signal at about 2 ppm. The pyrazole analogue (7) resonated significantly further upfield (~4 ppm, see Supporting Information) than did 5. These results suggest a critical role of the additional carbon between the nitrogens in imidazoles for balancing the pK_a of the exchangeable proton and the hydrogen bonding. Furthermore, the additional hydrogen bonds in I45DC were reported to stabilize its ‘folded’ conformation by 14 kcal/mol in solution and in the solid state, which also contribute to the salutary CEST contrast properties of 5 (60,61).

We then investigated the CEST properties of 5 in detail. As shown in Fig. 3, at fixed saturation field strengths (ω₁, with ω₁ = 5.9 µT shown), the CEST contrast changed significantly with pH. At pH 5.4, there was minimal contrast observed at 7.8 ppm despite the presence of significant amide CEST signal at 1.5 ppm. The contrast at 7.8 ppm increased, reaching a maximum at physiologic pH: 7% at pH 6.0, 13% at pH 6.6, 18% at pH 7.1 and 22% at pH 7.5. Above pH 7.5 the contrast decreased (see Supporting Information). This behaviour is in agreement with the acid–base properties of I45DCs. I45DCs were reported as low in basicity and exist as neutral species at pH 5–8. The imidazolium pK_a in I45DCs was reported to be around 1.5 (60).

Figure 3.

Experimental data for 5. (A) Z spectra; (B) MTR_asym spectra; (C) power dependence; (D) pH dependence of k_aw. Experimental conditions: CEST data were obtained at 25 mM concentration, pH 5.4, 6.1, 6.6, 7.1, 7.5, t_sat = 3 s, ω₁ = 5.9 µT and 37 °C. For power and pH dependence measurements, ω₁ = 1.2, 2.4, 3.6, 4.8, 7.2, 10.8 and 14.4 µT (see Supporting Information for additional data with more pH values).

To understand further the exchange mechanism of these compounds, the imidazole N–H k_aw of 5 was determined by performing the QUESP experiment at nearly physiological pH values, as shown in Fig. 3 (36), with contrast increasing as a function of saturation field strength. At pH values of 5.4 and 6.0, the N–Hexchanged too rapidly to achieve significant signal, roughly estimated as 35 100/s and 16 100/s, respectively. The rate reduced with increasing pH and reached a minimum of 5300/s at pH7.5. These large changes resulted in pH-dependent CEST contrast. The contrast at 7.8 ppm changed from 2 to 10% from pH6.0 to 7.5 at 3.6 µT, 8% to 22% at 5.4 µT, and 10% to 27% at 7.2 µT, as shown in Fig. 3C. The large chemical shift of the labile protons, pH sensitivity and high water solubility of 5 make it a suitable as a pH sensor. The mechanism of contrast enhancement is expected to be similar to the behaviour seen previously and attributed to changes in conformation as the pH drops (60), although an additional factor impacting proton exchange includes an increase in available protons.

In order to test if the I45DC scaffold could tolerate chemical modification, we prepared two additional water soluble analogues. The first, 4,5-bis[(Asp)carbonyl]-1H-imidazole (8), has a very similar structure to 5. The second, 4,5-bis[(Lys)carbonyl]-1H-imidazole (9), departs somewhat from 5 by containing a free amine on the side chains. As shown in Fig. 4, 8 worked similarly to 5, with a slightly slower k_aw (5000/s) at pH 7.4. On the other hand, 9 showed significant contrast at pH 6.2 but then contrast decreased with increasing pH, with k_aw increasing to 12 000/s at pH 7.2. The results indicated that CEST contrast of I45DCs can be generalized to similar structures and that the pH sensitivity can be finely tuned by modifying the peptide side chains.

Figure 4.

pH dependence of 8 and 9. Experimental conditions: CEST data were obtained at 25 mM concentration; Z spectra shown are at pH 5–8, t_sat = 3 s, ω₁ = 5.9 µT, 37 °C.

For pH sensing it is important to develop probes with signals that can be internally calibrated to take into account changes in agent concentration versus changes in pH. To take advantage of the strong but relatively pH inert (for pH 6.5–7.8) CEST signal region between 3 and 5 ppm (Fig. 3), we introduced a ratiometric method by dividing the pH sensitive signal from 7.2 – 7.8 ppm by the pH insensitive signal from 4.2 – 4.8 ppm based on the assignments in the Supporting Information. We first were interested in determining a suitable saturation field strength, and as shown in Supporting Information (Fig. S3) the contrast increased with saturation field strength well past the maximum we can achieve for T_sat = 3 s continuous wave saturation pulses using a 72 mm body transmit coil (ω₁ = 5.9 µT) for both 5 and 8. Compound 5 has a faster change in ratio(MTR_asym) with pH than compound 8 (Fig. S3B). Based on this, phantoms of 5 were prepared at 6.25 mM (pH 5.7, 6.1, 6.6, 6.8, 7.1, 7.4), 25 mM (pH 5.7, 6.2, 6.5, 6.8, 7.4) and 50 mM (pH 5.5, 6.0, 6.6, 7.3), and were scanned using ω₁ = 5.9 µT. The CEST contrast images and pH ratiometric images are shown for comparison in Fig. 5. The results indicate that the calibrated pH images agree well with the values determined via pH electrode over a range of different concentrations. Mean-while, the image contrasts for the tubes were significantly different from each other, with contrast decreasing with decreasing pH unidirectionally over the range of interest. This consistent result encouraged us to investigate I45DCs for in vivo applications.

Figure 5.

pH calibration of 5. Experimental conditions: CEST data were obtained at 6.25 mM, 25 mM or 50 mM concentration, t_sat = 3 s, ω₁ = 5.9 µT and 37 °C, with physiological pH values shown in the figure. (A) CEST contrast observed at 7.2–7.8 ppm (upper panel) and 4.2–4.8 ppm (lower panel) with different concentrations and pH values; (B) calibration plot using all tubes in the phantom (upper panel) and pH map after ratiometric calibration (lower panel); (C) experimental versus calculated pH. pH measurements were made with a precision of ±0.1 unit.

Next, we evaluated whether 5 or 8 could be detected after administration into live animals through injecting 100 µL of a 0.25 M solution of each into three mice and collecting CEST images. Images consisting of a single axial slice containing both kidneys were chosen (Fig. 6). We used a 12-point collection scheme (±4.2, 4.5, 4.8, 7.2, 7.5 and 7.8 ppm) to minimize experimental error and increase reproducibility, using ω₁ = 5.9 µT to reduce the sensitivity to B₀ inhomogeneity. The CEST signal at 7.5 ppm increased in both the calyx and cortex and reached a maximum at about 45 min post-injection (Fig. 6A, B), which indicated maximal probe uptake. The average CEST contrast was 7.6 ± 1.0% and 6.2 ± 1.0% over the whole kidney for compounds 5 and 8, respectively (Fig. 6D). Based on 5 (a) displaying a larger increase in ratio (MTR_asym) as a function of pH than 8, and (b) producing stronger CEST contrast in the kidneys after tail vein administration, it was selected for pH mapping. At the peak CEST contrast time point (45 min), the ratiometric method was applied to obtain a pH image and is shown in Fig. 6C. The average pH for the whole kidney was about 6.5 ± 0.1 (N = 3), which is consistent with the pH ~ 6.6–6.7 reported previously (54,62).

Figure 6.

In vivo kidney images and pH maps. Experimental conditions: t_sat = 3 s, ω₁ = 5.9 µT. (A) 7.5 ppm CEST contrast maps at pre-, post-35 min, post-45 min and post-55 min injection of compound 5; (B) 4.5 ppm CEST contrast maps for the same times as in A; (C) T₂w image and pH map using compound 5 for the same mouse; (D) average contrast over the whole kidney for pre- and post-45 min injection of compounds 5 and 8.

At this point, a number of diaCEST imaging probes have been developed that appear to be quite promising for pH mapping, including iodinated agents such as iopamidol, iobitridol, iopromide and our new I45DC compounds. One of the advantages of our I45DC scaffold is that the chemical shifts are the largest of all probes developed to date, which is an important factor to consider for studies on 3 T scanners, although on high field preclinical scanners the contrast produced was quite similar to that produced by iopamidol (3). Another advantage of the I45DC probes is the ease of modification through the schemes presented in this article, allowing adjustment of the exchange properties and biodistribution for a particular medical application. On the other hand, while some toxicity data for I45DC compounds are available in the literature, with the LD50 reported at 200–350 mg/kg in rodents (63–65), a clear advantage of the aforementioned iodinated CEST agents is their prior approval for administration to patients at high concentrations, which has allowed pilot studies of pH imaging technology in patients (62). Another disadvantage of the current group of I45DCs is that, with the largest chemical shift around 7.8 ppm, the k_ex values (~4000/s) are still not ideal for 3 T scanners (57). Based on these considerations, we believe the I45DC agents represent a strong new alternative to the existing iodinated pH imaging probes.

3. CONCLUSIONS

We have demonstrated that imidazole N–H protons can be tuned for CEST imaging through appropriate ring substitution. In particular I45DCs 5, 8 and 9 had imidazole N–H protons that resonated at 7.8 ppm from water, which produced significant contrast. Their CEST properties were systematically studied, with the contrast response to pH found to be significantly different upon attachment of lysine, suggesting their viability as pH sensors. Compound 5 was injected into mice and the pH of the kidneys was accurately measured.

4. EXPERIMENTAL

4.1. General for the tested compounds

Imidazole-5-carboxylic acid, imidazole-4,5-dicarboxylic acid, 1,2,3-triazole-4,5-dicarboxylic acid and pyrazole-3,5-dicarboxylic acid were purchased from Sigma Aldrich (St. Louis, MO, USA). H-Glu(Ot-Bu)Ot-Bu HCl, H-Asp(Ot-Bu)Ot-Bu HCl and H-Lys(N-Boc) Ot-Bu were purchased from Chem-Impex (Wood Dale, IL, USA). Solvents used for chromatography were ACS or HPLC grade, as appropriate. Reactions were monitored by thin layer chromatography (TLC) and by ¹H NMR. EM Science 60 F silica gel plates were used for TLC analyses. Flash column chromatography was performed over ICN EcoChrom silica gel (32–63 mm). HPLC purifications were performed using 18C columns (19 mm × 250 mm, Waters Atlantis) at 10 mL/min flow rate of water–acetonitrile eluent, unless specified otherwise. ¹H and ¹³C NMR spectra were recorded on a 400 MHz spectrometer. The chemical shifts are reported as δ values (ppm) relative to the water signal in D₂O. High-resolution mass spectral analyses were performed using the electrospray ionization (ESI) method.

4.2. Synthesis of (S)-2-(1H-imidazole-5-carboxamido)pentanedioic acid (3)

Diimidazo[1,5-a]piperazine-5,10-dione (10) (66) 376 mg (2 mmol) was dissolved in 20 mL dry THF. H-Glu(Ot-Bu)Ot-Bu HCl (11) 1.2 g (4 mmol) and triethyl amine 3 mL (20 mmol) were added to the solution at 0 °C and the reaction was stirred overnight at room temperature. After the solvent was removed under vacuum, the tert-butyl protected intermediate was obtained by flash column chromatography. Then, this intermediate was dissolved in 5 mL TFA/DCM (1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, compound 3 was purified by HPLC as a white powder, 310 mg, yield 23%.

¹H NMR (400 MHz, D₂O): δ 8.69 (s, 1H), 7.91 (s, 1H), 4.47 (dd, J₁ = 7.2 Hz, J₂ = 3.9 Hz, 1H), 2.38 (t, J = 5.4 Hz, 2H), 2.20–2.11 (m, 1H), 2.01–1.94 (m, 1H). ¹³C NMR (100 MHz, D₂O): δ 177.0, 174.6, 162.8 (q, TFA), 158.7, 135.5, 126.6, 120.7, 116.4 (q, TFA), 52.4, 29.8, 25.6. MS: HR-ESI calculated for (M + H⁺) C9H12N3O5⁺: 242.0771; found 242.0781.

HPLC (Waters Atlantis, MeCN/H₂O 8/92, 10 mL/min): 9.5 min.

4.3. Synthesis of imidazole-4,5-dicarboxamides (5, 8, 9)

5,10-Dioxo-5H,10H-diimidazo[1,5-a:1′-5′-d]pyrazine-1,6-dicarboxylic acid diphenyl ester (12) (67), 214 mg (0.5 mmol) and 5 mL THF were added to a dry flask. To this suspension at 0 °C was added the protected amino acids (11, 13 or 14; 1 mmol) and EtNi-Pr₂ 2 mL (11 mmol). After stirring at room temperature for 2 h, the reaction was refluxed for 2–4 days, monitored by TLC. After the solvent was removed under vacuum, the tert-butyl protected intermediate was obtained by flash column chromatography. Then, this intermediate was dissolved in 5 mL TFA/DCM (1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, compound 5, 8 or 9 was purified by HPLC.

5 yield 66% white powder.

¹H NMR (400 MHz, D₂O): δ 7.89 (s, 1H), 4.57 (dd, J₁ = 6.6 Hz, J₂ = 3.6 Hz, 2H), 2.45 (t, J = 5.4 Hz, 4H), 2.25–2.22 (m, 2H), 2.09–2.05 (m, 2H). ¹³C NMR (100 MHz, D₂O): δ 176.9, 174.5, 161.4, 136.9, 129.8, 52.1, 29.9, 26.0. MS: HR-ESI calculated for (M + Na⁺) C15H18N4NaO10⁺: 437.0915, found 437.0916.

HPLC (Waters Atlantis, MeCN/H₂O 15/85, 6 mL/min): 15.0 min.

8 yield 58% white powder.

¹H NMR (400 MHz, D₂O): δ 7.84 (s, 1H), 4.87 (t, J = 3.9 Hz, 2H), 3.05–2.91 (m, 4H). ¹³C NMR (100 MHz, D₂O): δ 174.3, 173.8, 161.2, 136.9, 129.8, 49.0, 35.6. MS: HR-ESI calculated for (M + Na⁺) C13H14N4NaO10⁺: 409.0602, found 409.0585.

HPLC (Waters Atlantis, MeCN/H₂O 15/85, 10 mL/min): 9.4 min.

9 yield 62% white powder.

¹H NMR (400 MHz, D₂O): δ 8.01 (s, 1H), 4.43 (dd, J₁ = 6.0 Hz, J₂ = 3.9 Hz, 2H), 2.86 (t, J = 5.4 Hz, 4H), 1.91–1.77 (m, 4H), 1.62–1.56 (m, 4H), 1.42–1.34 (m, 4H). ¹³C NMR (100 MHz, D₂O): δ 174.9, 162.7 (q, TFA), 160.7, 136.5, 129.1, 116.1 (q, TFA), 54.3, 52.9, 39.1, 30.0, 26.2, 21.9. MS: HR-ESI calculated for (M + H⁺) C17H29N6O6⁺: 413.2143, found 413.2130.

HPLC (Waters Atlantis, MeCN/H₂O 8/92, 10 mL/min): 15 min.

4.4. Synthesis of 6

1H-1,2,3-Triazole-4,5-dicarbonyl dichloride (16) was prepared using the same procedure as for 15 (67). 16 326 mg (1.7 mmol) and THF 10 mL were added to a dry flask. To this suspension at 0 °C was added H-Glu(Ot-Bu)Ot-Bu HCl (11) 1.0 g (3.4 mmol) and triethyl amine 3 mL (20 mmol). The reaction was stirred overnight at room temperature. After all of the solvent was removed under vacuum, the tert-butyl protected intermediate was obtained by flash column chromatography. Then, this intermediate was dissolved in 5 mL TFA/DCM (1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, Compound 6 was purified by HPLC as a light yellow powder, 300 mg, yield 43%.

¹H NMR (400 MHz, D₂O): δ 4.54 (dd, J₁ = 6.3 Hz, J₂ = 3.6 Hz, 2H), 2.39 (t, J = 5.4 Hz, 4H), 2.22–2.14 (m, 2H), 2.05–1.98 (m, 2H). ¹³C NMR (100 MHz, D₂O): δ 176.9, 174.2, 160.4, 137.1, 52.1, 29.8, 25.8. MS: HR-ESI calculated for (M + Na⁺) C15H18N4NaO10⁺: 437.0915, found 437.0921.

HPLC (Waters Atlantis, MeCN/H₂O 15/85, 10 mL/min): 12.5 min.

4.5. Synthesis of 7

1H-Pyrazole-3,5-dicarbonyl dichloride (15) (68) 384 mg (2 mmol) and THF 10 mL were added to a dry flask. To this suspension at 0 °C was added H-Glu(Ot-Bu)Ot-Bu HCl (11) 1.2 g (4 mmol) and triethyl amine 3 mL (20 mmol). The reaction was stirred overnight at room temperature. After all of the solvent was removed under vacuum, the tert-butyl protected intermediate was obtained by flash column chromatography. Then, this intermediate was dissolved in 5 mL TFA/DCM(1/1) for 2 h at room temperature. After all of the solvent was removed under vacuum, Compound 7 was purified by HPLC as a white powder, 280 mg, yield 34%.

¹H NMR (400 MHz, D₂O): δ 7.09 (s, 1H), 4.45 (dd, J₁ = 6.9 Hz, J₂ = 3.9 Hz, 2H), 2.38 (t, J = 5.4 Hz, 4H), 2.18–2.09 (m, 2H), 2.00–1.92 (m, 2H). ¹³C NMR (100 MHz, D₂O): δ 176.9, 174.6, 161.8, 141.6, 106.3, 52.0, 30.0, 25.6. MS: HR-ESI calculated for (M + Na⁺) C14H17N5NaO10⁺: 438.0868, found 438.0873.

HPLC (Waters Atlantis, MeCN/H₂O 15/85, 10 mL/min): 10.5 min.

4.6. Phantom solutions

Samples were dissolved in 0.01 M phosphate-buffered saline (PBS) at the desired concentrations and titrated using high concentration HCl/NaOH to the pH values listed. The solutions were then placed in 1 mm glass capillaries and assembled in a holder for CEST MR imaging. The samples were kept at 37 °C during imaging. Phantom CEST experiments were performed on a Bruker 11.7T vertical MR scanner, using a 20 mm birdcage transmit–receive coil. The CEST images were acquired using a RARE (RARE = 8) sequence with a CW saturation pulse length of 3 s and saturation field strengths (B₁) from 1.2 µT to 14.4 µT. The CEST Z spectra were acquired by incrementing the saturation frequency every 0.3 ppm from −15 to 15 ppm for phantoms; TR = 6 s, effective TE = 17 ms, matrix size = 64 × 48 and slice thickness of 1.2 mm.

4.7. Animal imaging

All experiments conducted with mice were performed in accordance with protocols approved by the Johns Hopkins University Institutional Animal Care and Use Committee (IACUC). BALB/c mice weighing 20–25 g (Charles River Laboratories) were maintained under specific pathogen free conditions in the Johns Hopkins University animal facility. During MRI acquisition, the mice were anesthetized using 0.5–2% isoflurane and placed in a 23 mm transmit–receive mouse coil. Their breath rate was monitored throughout the in vivo MRI experiments using a respiratory probe. To minimize motion artifacts, we selected a 1.5 mm thick slice in the middle of the calyx and kept the respiratory rate constant at about 30 breaths/min. A 100 µL volume of a 0.25 M I45DC solution in PBS (pH 7.3) was slowly injected via a catheter into the tail vein. In vivo images were acquired on a Bruker BioSpec 11.7 T horizontal MR scanner, with one axial slice 1.5 mm thick crossing both renal centres chosen for the CEST measurements. CEST images with saturation frequencies of (±7.2 ppm, ±7.5 ppm, ±7.8 ppm) and (±4.2 ppm, ±4.5 ppm, ±4.8 ppm) were acquired repeatedly every 10 min both pre- and post-injection. Another set of saturation weighted images with frequency incrementing every 0.1 ppm from −1 to 1 ppm, termed the water saturation shift reference (WASSR) (69), were collected for B₀ mapping as the first pre-injection and the final post-injection set of images, using a 0.5 s saturation pulse with B₁ = 0.5 µT, and TR/effective TE = 2 s/15 ms. Image parameters were similar to those for the phantom except for TR/effective TE = 5 s/15 ms, and setting B₁ = 5.9 µT. CEST contrast was quantified as in phantoms.

4.8. Post-processing

All post-processing was performed using in-house written MATLAB scripts. CEST contrast was quantified in the images using

$$\mathrm{M}\mathrm{T}{\mathrm{R}}_{\text{asym}}=[S(-\mathrm{\Delta }\omega )-S(+\mathrm{\Delta }\omega )]/S(-\mathrm{\Delta }\omega ).$$ (1)

With this metric based on the subtraction of the two water signal intensities with saturation pulses at frequencies of +Δω and −Δω with respect to water, i.e. S(+Δω) and S(−Δω), a voxel-by-voxel B₀ map was generated as described previously (69) using WASSR images for phantoms and live animals. As is shown for a representative mouse in the supporting information (Fig. S1), −75 Hz < B₀ < +75 Hz for the kidneys, which is much smaller than the magnitude of the saturation pulse strength (ω₁ = 5.9 µT). Because of this, B₀ correction was not necessary to perform in vivo. Instead, average maps of MTR_asym using the 12 offsets were calculated. All images were masked due to visceral movements present in regions outside the kidney; however, representative unmasked images are displayed in the supporting information (Fig. S2).

4.9. pH calibration using phantoms and in vivo pH map generation

The pH calibration scheme was developed based on examining the data for 5 at three different concentrations (6.25 mM, 25 mM and 50 mM), with each concentration containing four or five samples with pH varied from 5.7 to 7.4. For all the 14 samples of each agent, the ratio between the average MTR_asym value of (±7.2 ppm, ±7.5 ppm, ±7.8 ppm) and that of (±4.2 ppm, ±4.5 ppm, ±4.8 ppm) (B₁ = 5.9 µT) was plotted as a function of measured pH, with the correlation function of pH = ratio × 1.04 + 5.21. Based on the ratio and pH correlation function from the phantoms, the in vivo dynamic pH maps of the kidneys were generated.